Hi everyone and thanks for reporting. It's late for me so I'll be quick in my answers ;)
> With this patch the support of SSE2 isn't detected for my CPU and the > compilation ends successfully and the binary of blender run. As I understand your patch, SSE2 is not disabled. You have just switched optimizations from -O2 to -O0 (c.f http://gcc.gnu.org/onlinedocs/gcc/Optimize-Options.html#index-O2-683) I think this bug is much more complex than switching SSE2 and I suspect gcc to mess up somewhere. By the way, there is already a function in blender, called BLI_cpu_support_sse2, that check SSE2 capabilities: ---snip--- % find -regex ".*\.\(h\|c\)" -exec grep -HnE "sse2" {} \; ./source/blender/blenlib/BLI_cpu.h:30:int BLI_cpu_support_sse2(void); ./source/blender/blenlib/intern/cpu.c:31:int BLI_cpu_support_sse2(void) ./source/blender/render/intern/source/rayshade.c:108: type = BLI_cpu_support_sse2()? R_RAYSTRUCTURE_SIMD_SVBVH: R_RAYSTRUCTURE_VBVH; ---snip--- > Reading previus e-mails, I suggest to create 2 packages of blender: > with and without SSE2 support I'm not sure creating 2 packages based on compilation option is a good solution and this will make blender harder to maintain. I didn't have myself a amd-based computer, so I will be really happy if could test the last 2.57 package in sid. I will forward the bug to upstream and ask them some help on this. Best Regards, --- Kevin Roy -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
